Display growing count of objects

[kurtgn (Mikhail Novikov)]

I am registering subscribers in my Elasticsearch database. Each subscriber has a CREATED_AT timestamp field. And I can create a chart that shows how many new subscribers I get each day. This is easily done just by grouping the documents into timestamp buckets.

But what if I need to show an ever-growing amount of users each day? For example:

• on June 1st I had total 15 users (counting all objects with CREATED_AT <= June 1st)
• on June 2nd I had 4 new users and a total of 19 users (counting all objects with CREATED_AT <= June 2nd)
• on June 3rd I had 10 new users and a total of 29 users (counting all objects with CREATED_AT <= June 3rd)
• etc

And this should result in a timeseries data like this:

June 1st: 15
June 2nd: 19
June 3rd: 29
...

What aggregation should I pick for this?

[abrx]

Just plot the numbers of documents (aka users) by timestamp (bar chart or plot or whatever) and set minimum interval to daily in the x-axis date histogram specifications.
It will show how many documents (filtered if necessary to get only new users) indexed each days.

[kurtgn (Mikhail Novikov)]

thanks, but which one of these aggregations is the number of documents by timestamp? I only see count, and this count is separate for each day, it does not include counts for all previous timestamps.

[abrx]

Count, as described here.

[kurtgn (Mikhail Novikov)]

Count does not work for me, because In my case, the count will show:

June 1st: 15
June 2nd: 4
June 3rd: 10
...

but I need something that would show an ever-incrementing value:

June 1st: 15
June 2nd: 19
June 3rd: 29
...

where each bucket includes counts for all previous days + its own count.